DailyAssignment21

# Library
library(tsibble)
library(feasts)
library(fable)
library(ggplot2)
library(plotly)
library(dplyr)
library(readr)

#Load dataset
set.seed(123)
my_data <- tibble(
  date = seq(as.Date("2015-01-01"), as.Date("2020-12-31"), by = "month"),
  streamflow = 100 + sin(2 * pi * 1:72 / 12) * 10 + rnorm(72, 0, 5)
)

# Convert to tsibble 
ts_data <- my_data %>%
  as_tsibble(index = date)

# Plot Time Series
p <- ggplot(ts_data, aes(x = date, y = streamflow)) +
  geom_line(color = "darkblue") +
  labs(title = "Monthly Streamflow Over Time", y = "Streamflow", x = "Date") +
  theme_minimal()

# Animate with Plotly
ggplotly(p)

library(dplyr)
library(tsibble)
library(imputeTS)
library(feasts)
library(ggplot2)

# Step 1: Rebuild tsibble and ensure time index is regular
ts_data <- my_data %>%
  as_tsibble(index = date) %>%
  fill_gaps()                           # Fills any missing time stamps

# Step 2: Interpolate missing values fully
ts_data <- ts_data %>%
  mutate(streamflow = na_interpolation(streamflow, option = "linear"))

# Confirm all NAs are gone
sum(is.na(ts_data$streamflow))   # MUST be 0

[1] 0

library(ggplot2)

ggplot(ts_data, aes(x = date, y = streamflow)) +
  geom_line(color = "forestgreen") +
  labs(title = "Streamflow Time Series", x = "Date", y = "Streamflow")

# Subseries Plot
library(tsibble)
library(dplyr)
library(lubridate)
library(feasts)


my_data <- my_data %>%
  mutate(date = as.Date(date))

# Step 2: Create a tsibble
ts_data <- my_data %>%
  as_tsibble(index = date) %>%
  fill_gaps()

# Interpolate missing values
library(imputeTS)
ts_data$streamflow <- na_interpolation(ts_data$streamflow, option = "linear")

# Aggregate to monthly streamflow
ts_monthly <- ts_data %>%
  mutate(month = yearmonth(date)) %>%   # This must come AFTER date is a Date
  index_by(month) %>%
  summarise(streamflow = mean(streamflow, na.rm = TRUE))

gg_subseries(ts_monthly, streamflow) +
  labs(title = "Monthly Subseries Plot of Streamflow",
       y = "Streamflow", x = "Month") +
  theme_minimal()

This sub-series plot shows streamflow values grouped by month (i.e., “seasons”). Peaks in certain months suggest consistent seasonal patterns. For instance, higher values in spring months may indicate snowmelt contributions to streamflow.

# Decomposition
decomp <- ts_data %>%
  model(STL(streamflow ~ season(window = "periodic"))) %>%
  components()

autoplot(decomp) +
  labs(title = "STL Decomposition of Streamflow")